Statistical evaluation of music information retrieval experiments
Author:
- 1 June 2006
- journal article
- research article
- Published by Taylor & Francis in Journal of New Music Research
- Vol. 35 (2) , 113-120
Abstract
This work concerns the necessity of statistical evaluation of Music Information Retrieval (MIR) experiments. This necessity is motivated by applying fundamental notions of statistical hypotheses testing to MIR research. Minimum requirements concerning statistical evaluation are developed and the appropriate statistical techniques are introduced and exemplified in a genre classification context. Articles from the MIR literature are examined and criticized for the lack of statistical evaluation they contain.Keywords
This publication has 10 references indexed in Scilit:
- The Scientific Evaluation of Music Information Retrieval Systems: Foundations and FutureComputer Music Journal, 2004
- Musical genre classification of audio signalsIEEE Transactions on Speech and Audio Processing, 2002
- The Elements of Statistical Learning2001
- A music similarity function based on signal analysis2001
- Approximate Statistical Tests for Comparing Supervised Classification Learning AlgorithmsNeural Computation, 1998
- On Comparing Classifiers: Pitfalls to Avoid and a Recommended ApproachData Mining and Knowledge Discovery, 1997
- A quantitative study of experimental evaluations of neural network learning algorithms: Current research practiceNeural Networks, 1996
- On the quality of neural net classifiersArtificial Intelligence in Medicine, 1994
- The Jackknife, the Bootstrap and Other Resampling Plans1982
- The Analysis of Contingency Tables1977